In 2008, over 66,000 people in the US will be diagnosed with Non-Hodgkin lymphoma (NHL), and over 19,000 will die of this cancer. Survival rates have only recently begun to improve, and the current 5-year survival rate is 66%. We have identified a number of candidate host (inherited) immune genes and DNA repair genes that individually and in aggregate predict survival for follicular lymphoma (FL) and diffuse large B-cell lymphoma (DLBCL) beyond classic clinical and demographic prognostic factors. We have also found that pre-diagnosis smoking and obesity are associated with poorer overall NHL survival. We propose to replicate and extend these provocative findings. Furthermore, building off these findings, we will evaluate whether the association of host genetics is independent of tumor molecular markers, as well as begin to explore whether there are host-tumor interactions that impact disease progression and survival. The overall goal of this study is to identify host genetic and tumor molecular markers that predict event-free and overall survival in order to improve prognostication, better understand NHL pathophysiology, and ultimately help identify approaches to improve the survival of NHL patients. Our specific aims are: 1) To evaluate the association of polymorphisms in immune and DNA repair genes with event-free and overall survival from FL and DLBCL;2) To evaluate the association of tumor molecular markers with event-free and overall survival from FL and DLBCL;and 3) To develop multivariate prediction models for FL and DLBCL that integrate standard demographic and clinical characteristics, treatment, host genetic variation (Aim 1) and tumor molecular markers (Aim 2) to predict eventfree and overall survival. In a secondary aim, we will evaluate the role of pre-diagnosis lifestyle factors with event-free and overall survival from NHL. To achieve these aims, we will use the Lymphoma Molecular Epidemiology Resource, an ongoing, prospective prognostic cohort study of newly diagnosed cases from the Mayo Clinic and the University of Iowa initiated in 2002. This resource has several methodologic strengths, including large sample size (>2200cases), central pathology review and classification, availability of tumor tissue, detailed baseline clinical prognostic data, initial and subsequent treatments, and a large patient population treated in the immunochemotherapy (rituximab) era. We have systematically collected detailed outcome data, which will allow us to evaluate both event-free and overall survival. We will also be able to evaluate host genetic and tumor molecular prognostic factors in the context of established demographic and clinical prognostic factors, as well as all treatment(s). Upon completion of these aims, we will have simultaneously evaluated the role of host genetic variation and molecular tumor markers in the prognosis of FL and DLBCL. If host genetic factors are confirmed as robust predictors of outcome, this could lead to a fundamental shift in how patient prognosis is evaluated by incorporating host genotype into prognostic models. Host genetics may also lead to a better understanding of NHL pathophysiology that could lead to new approaches to improve the survival of NHL patients.